Latent information in the pattem of missing observations in global mail surveys

Incomplete responses are quite common in global mail surveys. A review of empirical research published in the last decade revealed that this problem is pervasive, and missing responses to the extent of 15% of total responses are not unusual. Most researchers use simple methods like deletion or substitution to deal with this problem. However, these methods ignore the basic philosophical assumptions of the field and do not consider the information hidden in the pattern of the missing observations. The exploratory study reported here analyzed two existing data sets which had been previously collected with mail survey to study technology transfer issues. Both data sets had a significant amount of missing responses. Our study results reveal that the pattern of missing responses was not random in nature, and there was a significant association between socio-demographic characteristics of the respondents and the number of missing responses.